Artificial Intelligence Nanodegree

Convolutional Neural Networks

Project: Write an Algorithm for a Dog Identification App


In this notebook, some template code has already been provided for you, and you will need to implement additional functionality to successfully complete this project. You will not need to modify the included code beyond what is requested. Sections that begin with '(IMPLEMENTATION)' in the header indicate that the following block of code will require additional functionality which you must provide. Instructions will be provided for each section, and the specifics of the implementation are marked in the code block with a 'TODO' statement. Please be sure to read the instructions carefully!

Note: Once you have completed all of the code implementations, you need to finalize your work by exporting the iPython Notebook as an HTML document. Before exporting the notebook to html, all of the code cells need to have been run so that reviewers can see the final implementation and output. You can then export the notebook by using the menu above and navigating to \n", "File -> Download as -> HTML (.html). Include the finished document along with this notebook as your submission.

In addition to implementing code, there will be questions that you must answer which relate to the project and your implementation. Each section where you will answer a question is preceded by a 'Question X' header. Carefully read each question and provide thorough answers in the following text boxes that begin with 'Answer:'. Your project submission will be evaluated based on your answers to each of the questions and the implementation you provide.

Note: Code and Markdown cells can be executed using the Shift + Enter keyboard shortcut. Markdown cells can be edited by double-clicking the cell to enter edit mode.

The rubric contains optional "Stand Out Suggestions" for enhancing the project beyond the minimum requirements. If you decide to pursue the "Stand Out Suggestions", you should include the code in this IPython notebook.


Why We're Here

In this notebook, you will make the first steps towards developing an algorithm that could be used as part of a mobile or web app. At the end of this project, your code will accept any user-supplied image as input. If a dog is detected in the image, it will provide an estimate of the dog's breed. If a human is detected, it will provide an estimate of the dog breed that is most resembling. The image below displays potential sample output of your finished project (... but we expect that each student's algorithm will behave differently!).

Sample Dog Output

In this real-world setting, you will need to piece together a series of models to perform different tasks; for instance, the algorithm that detects humans in an image will be different from the CNN that infers dog breed. There are many points of possible failure, and no perfect algorithm exists. Your imperfect solution will nonetheless create a fun user experience!

The Road Ahead

We break the notebook into separate steps. Feel free to use the links below to navigate the notebook.

  • Step 0: Import Datasets
  • Step 1: Detect Humans
  • Step 2: Detect Dogs
  • Step 3: Create a CNN to Classify Dog Breeds (from Scratch)
  • Step 4: Use a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)
  • Step 6: Write your Algorithm
  • Step 7: Test Your Algorithm

Step 0: Import Datasets

Import Dog Dataset

In the code cell below, we import a dataset of dog images. We populate a few variables through the use of the load_files function from the scikit-learn library:

  • train_files, valid_files, test_files - numpy arrays containing file paths to images
  • train_targets, valid_targets, test_targets - numpy arrays containing onehot-encoded classification labels
  • dog_names - list of string-valued dog breed names for translating labels
In [1]:
from sklearn.datasets import load_files       
from keras.utils import np_utils
import numpy as np
from glob import glob

# define function to load train, test, and validation datasets
def load_dataset(path):
    data = load_files(path)
    dog_files = np.array(data['filenames'])
    dog_targets = np_utils.to_categorical(np.array(data['target']), 133)
    return dog_files, dog_targets

# load train, test, and validation datasets
train_files, train_targets = load_dataset('dogImages/train')
valid_files, valid_targets = load_dataset('dogImages/valid')
test_files, test_targets = load_dataset('dogImages/test')

# load list of dog names
dog_names = [item[20:-1] for item in sorted(glob("dogImages/train/*/"))]

# print statistics about the dataset
print('There are %d total dog categories.' % len(dog_names))
print('There are %s total dog images.\n' % len(np.hstack([train_files, valid_files, test_files])))
print('There are %d training dog images.' % len(train_files))
print('There are %d validation dog images.' % len(valid_files))
print('There are %d test dog images.'% len(test_files))
Using TensorFlow backend.
/home/yp/Applications/anaconda3/envs/dlnf/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6
  return f(*args, **kwds)
There are 133 total dog categories.
There are 8351 total dog images.

There are 6680 training dog images.
There are 835 validation dog images.
There are 836 test dog images.

Import Human Dataset

In the code cell below, we import a dataset of human images, where the file paths are stored in the numpy array human_files.

In [2]:
import random
random.seed(8675309)

# load filenames in shuffled human dataset
human_files = np.array(glob("lfw/*/*"))
random.shuffle(human_files)

# print statistics about the dataset
print('There are %d total human images.' % len(human_files))
There are 13233 total human images.

Step 1: Detect Humans

We use OpenCV's implementation of Haar feature-based cascade classifiers to detect human faces in images. OpenCV provides many pre-trained face detectors, stored as XML files on github. We have downloaded one of these detectors and stored it in the haarcascades directory.

In the next code cell, we demonstrate how to use this detector to find human faces in a sample image.

In [3]:
import cv2                
import matplotlib.pyplot as plt                        
%matplotlib inline                               

# extract pre-trained face detector
face_cascade = cv2.CascadeClassifier('haarcascades/haarcascade_frontalface_alt.xml')

# load color (BGR) image
img = cv2.imread(human_files[3])
# convert BGR image to grayscale
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)

# find faces in image
faces = face_cascade.detectMultiScale(gray)

# print number of faces detected in the image
print('Number of faces detected:', len(faces))

# get bounding box for each detected face
for (x,y,w,h) in faces:
    # add bounding box to color image
    cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)
    
# convert BGR image to RGB for plotting
cv_rgb = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)

# display the image, along with bounding box
plt.imshow(cv_rgb)
plt.show()
Number of faces detected: 1

Before using any of the face detectors, it is standard procedure to convert the images to grayscale. The detectMultiScale function executes the classifier stored in face_cascade and takes the grayscale image as a parameter.

In the above code, faces is a numpy array of detected faces, where each row corresponds to a detected face. Each detected face is a 1D array with four entries that specifies the bounding box of the detected face. The first two entries in the array (extracted in the above code as x and y) specify the horizontal and vertical positions of the top left corner of the bounding box. The last two entries in the array (extracted here as w and h) specify the width and height of the box.

Write a Human Face Detector

We can use this procedure to write a function that returns True if a human face is detected in an image and False otherwise. This function, aptly named face_detector, takes a string-valued file path to an image as input and appears in the code block below.

In [4]:
# returns "True" if face is detected in image stored at img_path
def face_detector(img_path):
    img = cv2.imread(img_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    faces = face_cascade.detectMultiScale(gray)
    return [len(faces) > 0, faces]

(IMPLEMENTATION) Assess the Human Face Detector

Question 1: Use the code cell below to test the performance of the face_detector function.

  • What percentage of the first 100 images in human_files have a detected human face?
  • What percentage of the first 100 images in dog_files have a detected human face?

Ideally, we would like 100% of human images with a detected face and 0% of dog images with a detected face. You will see that our algorithm falls short of this goal, but still gives acceptable performance. We extract the file paths for the first 100 images from each of the datasets and store them in the numpy arrays human_files_short and dog_files_short.

Answer:

  • 97% is detected in human files.
  • 11% is detected in dog files.
In [59]:
human_files_short = human_files[:100]
dog_files_short = train_files[:100]
# Do NOT modify the code above this line.

## TODO: Test the performance of the face_detector algorithm 
## on the images in human_files_short and dog_files_short.
# display the image, along with bounding box


# in human files
pos_idx = {}
total_cnt = 0
for idx, img_path in enumerate(human_files_short):
    pos, faces = face_detector(img_path)
    if pos == True:
        pos_idx[idx] = faces
    else:
        img = cv2.imread(img_path)
        plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
        plt.title('undetected')
        plt.show()
    total_cnt += 1
print('{}% is detected in humans.'.format(len(pos_idx) / total_cnt * 100))

fig = plt.figure(figsize=(32, 32))
img_i = 1
cols = 10
rows = 10

# show detected images
for idx in pos_idx.keys():
    img = cv2.imread(human_files_short[idx])
    fig.add_subplot(rows, cols, img_i)
    img_i += 1

    # get bounding box for each detected face
    faces = pos_idx[idx]
    for (x,y,w,h) in faces:
        # add bounding box to color image
        cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)

    plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
plt.show()

# in dog files
fig = plt.figure(figsize=(32, 32))
img_i = 1
cols = 10
rows = 10

pos_idx.clear()
total_cnt = 0
for idx, img_path in enumerate(dog_files_short):
    pos, faces = face_detector(img_path)
    if pos == True:
        pos_idx[idx] = faces
    total_cnt += 1
print('{}% is detected in dogs.'.format(len(pos_idx) / total_cnt * 100))

# show detected images
for idx in pos_idx.keys():
    img = cv2.imread(dog_files_short[idx])
    fig.add_subplot(rows, cols, img_i)
    img_i += 1

    # get bounding box for each detected face
    faces = pos_idx[idx]
    for (x,y,w,h) in faces:
        # add bounding box to color image
        cv2.rectangle(img,(x,y),(x+w,y+h),(255,0,0),2)

    plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
plt.show()
97.0% is detected in humans.
11.0% is detected in dogs.

Question 2: This algorithmic choice necessitates that we communicate to the user that we accept human images only when they provide a clear view of a face (otherwise, we risk having unneccessarily frustrated users!). In your opinion, is this a reasonable expectation to pose on the user? If not, can you think of a way to detect humans in images that does not necessitate an image with a clearly presented face?

Answer: It depends on the necessity of an application. It will be worth when the application is interactive like a locking system using human face. But if the application is a monitoring or detecting system which cannot be interactive with human, this will be a huge weakness. To detect humans, I can think several ways such as detecting face, hair, clothes, shape, and so on. So, building up a neural network would return comprehensive result for detecting humans.

We suggest the face detector from OpenCV as a potential way to detect human images in your algorithm, but you are free to explore other approaches, especially approaches that make use of deep learning :). Please use the code cell below to design and test your own face detection algorithm. If you decide to pursue this optional task, report performance on each of the datasets.

In [6]:
## (Optional) TODO: Report the performance of another  
## face detection algorithm on the LFW dataset
### Feel free to use as many code cells as needed.

Step 2: Detect Dogs

In this section, we use a pre-trained ResNet-50 model to detect dogs in images. Our first line of code downloads the ResNet-50 model, along with weights that have been trained on ImageNet, a very large, very popular dataset used for image classification and other vision tasks. ImageNet contains over 10 million URLs, each linking to an image containing an object from one of 1000 categories. Given an image, this pre-trained ResNet-50 model returns a prediction (derived from the available categories in ImageNet) for the object that is contained in the image.

In [7]:
from keras.applications.resnet50 import ResNet50

# define ResNet50 model
ResNet50_model = ResNet50(weights='imagenet')

Pre-process the Data

When using TensorFlow as backend, Keras CNNs require a 4D array (which we'll also refer to as a 4D tensor) as input, with shape

$$ (\text{nb_samples}, \text{rows}, \text{columns}, \text{channels}), $$

where nb_samples corresponds to the total number of images (or samples), and rows, columns, and channels correspond to the number of rows, columns, and channels for each image, respectively.

The path_to_tensor function below takes a string-valued file path to a color image as input and returns a 4D tensor suitable for supplying to a Keras CNN. The function first loads the image and resizes it to a square image that is $224 \times 224$ pixels. Next, the image is converted to an array, which is then resized to a 4D tensor. In this case, since we are working with color images, each image has three channels. Likewise, since we are processing a single image (or sample), the returned tensor will always have shape

$$ (1, 224, 224, 3). $$

The paths_to_tensor function takes a numpy array of string-valued image paths as input and returns a 4D tensor with shape

$$ (\text{nb_samples}, 224, 224, 3). $$

Here, nb_samples is the number of samples, or number of images, in the supplied array of image paths. It is best to think of nb_samples as the number of 3D tensors (where each 3D tensor corresponds to a different image) in your dataset!

In [8]:
from keras.preprocessing import image                  
from tqdm import tqdm

def path_to_tensor(img_path):
    # loads RGB image as PIL.Image.Image type
    img = image.load_img(img_path, target_size=(224, 224))
    # convert PIL.Image.Image type to 3D tensor with shape (224, 224, 3)
    x = image.img_to_array(img)
    # convert 3D tensor to 4D tensor with shape (1, 224, 224, 3) and return 4D tensor
    return np.expand_dims(x, axis=0)

def paths_to_tensor(img_paths):
    list_of_tensors = [path_to_tensor(img_path) for img_path in tqdm(img_paths)]
    return np.vstack(list_of_tensors)

Making Predictions with ResNet-50

Getting the 4D tensor ready for ResNet-50, and for any other pre-trained model in Keras, requires some additional processing. First, the RGB image is converted to BGR by reordering the channels. All pre-trained models have the additional normalization step that the mean pixel (expressed in RGB as $[103.939, 116.779, 123.68]$ and calculated from all pixels in all images in ImageNet) must be subtracted from every pixel in each image. This is implemented in the imported function preprocess_input. If you're curious, you can check the code for preprocess_input here.

Now that we have a way to format our image for supplying to ResNet-50, we are now ready to use the model to extract the predictions. This is accomplished with the predict method, which returns an array whose $i$-th entry is the model's predicted probability that the image belongs to the $i$-th ImageNet category. This is implemented in the ResNet50_predict_labels function below.

By taking the argmax of the predicted probability vector, we obtain an integer corresponding to the model's predicted object class, which we can identify with an object category through the use of this dictionary.

In [9]:
from keras.applications.resnet50 import preprocess_input, decode_predictions

def ResNet50_predict_labels(img_path):
    # returns prediction vector for image located at img_path
    img = preprocess_input(path_to_tensor(img_path))
    return np.argmax(ResNet50_model.predict(img))

Write a Dog Detector

While looking at the dictionary, you will notice that the categories corresponding to dogs appear in an uninterrupted sequence and correspond to dictionary keys 151-268, inclusive, to include all categories from 'Chihuahua' to 'Mexican hairless'. Thus, in order to check to see if an image is predicted to contain a dog by the pre-trained ResNet-50 model, we need only check if the ResNet50_predict_labels function above returns a value between 151 and 268 (inclusive).

We use these ideas to complete the dog_detector function below, which returns True if a dog is detected in an image (and False if not).

In [10]:
### returns "True" if a dog is detected in the image stored at img_path
def dog_detector(img_path):
    prediction = ResNet50_predict_labels(img_path)
    return [((prediction <= 268) & (prediction >= 151)), prediction]

(IMPLEMENTATION) Assess the Dog Detector

Question 3: Use the code cell below to test the performance of your dog_detector function.

  • What percentage of the images in human_files_short have a detected dog?
  • What percentage of the images in dog_files_short have a detected dog?

Answer:

  • 1% is detected in human files.
  • 100% is detected in dog files.
In [11]:
### TODO: Test the performance of the dog_detector function
### on the images in human_files_short and dog_files_short.

# in human files
pos_idx = {}
total_cnt = 0
for idx, img_path in enumerate(human_files_short):
    pos, prediction = dog_detector(img_path)
    if pos == True:
        pos_idx[idx] = prediction
    total_cnt += 1
print('{}% is detected in humans.'.format(len(pos_idx) / total_cnt * 100))

# show detected images
for idx, prediction in pos_idx.items():
    img = cv2.imread(human_files_short[idx])
    plt.imshow(cv2.cvtColor(img, cv2.COLOR_BGR2RGB))
    plt.show()
    print('prediction =', prediction)

# in dog files
pos_idx.clear()
total_cnt = 0
for idx, img_path in enumerate(dog_files_short):
    pos, prediction = dog_detector(img_path)
    if pos == True:
        pos_idx[idx] = img_path
    total_cnt += 1
print('{}% is detected in dogs.'.format(len(pos_idx) / total_cnt * 100))
1.0% is detected in humans.
prediction = 255
100.0% is detected in dogs.

Step 3: Create a CNN to Classify Dog Breeds (from Scratch)

Now that we have functions for detecting humans and dogs in images, we need a way to predict breed from images. In this step, you will create a CNN that classifies dog breeds. You must create your CNN from scratch (so, you can't use transfer learning yet!), and you must attain a test accuracy of at least 1%. In Step 5 of this notebook, you will have the opportunity to use transfer learning to create a CNN that attains greatly improved accuracy.

Be careful with adding too many trainable layers! More parameters means longer training, which means you are more likely to need a GPU to accelerate the training process. Thankfully, Keras provides a handy estimate of the time that each epoch is likely to take; you can extrapolate this estimate to figure out how long it will take for your algorithm to train.

We mention that the task of assigning breed to dogs from images is considered exceptionally challenging. To see why, consider that even a human would have great difficulty in distinguishing between a Brittany and a Welsh Springer Spaniel.

Brittany Welsh Springer Spaniel

It is not difficult to find other dog breed pairs with minimal inter-class variation (for instance, Curly-Coated Retrievers and American Water Spaniels).

Curly-Coated Retriever American Water Spaniel

Likewise, recall that labradors come in yellow, chocolate, and black. Your vision-based algorithm will have to conquer this high intra-class variation to determine how to classify all of these different shades as the same breed.

Yellow Labrador Chocolate Labrador Black Labrador

We also mention that random chance presents an exceptionally low bar: setting aside the fact that the classes are slightly imabalanced, a random guess will provide a correct answer roughly 1 in 133 times, which corresponds to an accuracy of less than 1%.

Remember that the practice is far ahead of the theory in deep learning. Experiment with many different architectures, and trust your intuition. And, of course, have fun!

Pre-process the Data

We rescale the images by dividing every pixel in every image by 255.

In [12]:
from PIL import ImageFile
ImageFile.LOAD_TRUNCATED_IMAGES = True

# pre-process the data for Keras
train_tensors = paths_to_tensor(train_files).astype('float32')/255
valid_tensors = paths_to_tensor(valid_files).astype('float32')/255
test_tensors = paths_to_tensor(test_files).astype('float32')/255
100%|██████████| 6680/6680 [00:33<00:00, 200.61it/s]
100%|██████████| 835/835 [00:09<00:00, 89.57it/s]
100%|██████████| 836/836 [00:09<00:00, 89.88it/s]

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:

    model.summary()

We have imported some Python modules to get you started, but feel free to import as many modules as you need. If you end up getting stuck, here's a hint that specifies a model that trains relatively fast on CPU and attains >1% test accuracy in 5 epochs:

Sample CNN

Question 4: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. If you chose to use the hinted architecture above, describe why you think that CNN architecture should work well for the image classification task.

Answer: I tried to extract features 4 times as 32, 64, 128, and 256 in order using 'Conv2D'. And between layers, I tried to maximize the difference using 'MaxPooling2D' respectively. For the memory reason, I used 'GlobalAveragePolling2D' to make a fully connected layer. Finally, to classify the breed of dog, I used 133 'Dense' layer.

In [13]:
from keras.layers import Conv2D, MaxPooling2D, GlobalAveragePooling2D
from keras.layers import Dropout, Flatten, Dense
from keras.models import Sequential

model = Sequential()

### TODO: Define your architecture.
model.add(Conv2D(filters=32, kernel_size=5, padding='same', activation='relu', input_shape=(224, 224, 3)))
model.add(MaxPooling2D(pool_size=2))
model.add(Conv2D(filters=64, kernel_size=5, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Conv2D(filters=128, kernel_size=5, padding='same', activation='relu'))
model.add(MaxPooling2D(pool_size=2))
model.add(Conv2D(filters=256, kernel_size=5, padding='same', activation='relu'))
model.add(GlobalAveragePooling2D())
model.add(Dropout(0.2))
model.add(Dense(133, activation='softmax'))

model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_1 (Conv2D)            (None, 224, 224, 32)      2432      
_________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 112, 112, 32)      0         
_________________________________________________________________
conv2d_2 (Conv2D)            (None, 112, 112, 64)      51264     
_________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 56, 56, 64)        0         
_________________________________________________________________
conv2d_3 (Conv2D)            (None, 56, 56, 128)       204928    
_________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 28, 28, 128)       0         
_________________________________________________________________
conv2d_4 (Conv2D)            (None, 28, 28, 256)       819456    
_________________________________________________________________
global_average_pooling2d_1 ( (None, 256)               0         
_________________________________________________________________
dropout_1 (Dropout)          (None, 256)               0         
_________________________________________________________________
dense_1 (Dense)              (None, 133)               34181     
=================================================================
Total params: 1,112,261
Trainable params: 1,112,261
Non-trainable params: 0
_________________________________________________________________

Compile the Model

In [14]:
model.compile(optimizer='rmsprop', loss='categorical_crossentropy', metrics=['accuracy'])

(IMPLEMENTATION) Train the Model

Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.

You are welcome to augment the training data, but this is not a requirement.

In [17]:
from keras.callbacks import ModelCheckpoint  

ImageFile.LOAD_TRUNCATED_IMAGES = True
### TODO: specify the number of epochs that you would like to use to train the model.

epochs = 50

### Do NOT modify the code below this line.

checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.from_scratch.hdf5', 
                               verbose=1, save_best_only=True)

history = model.fit(train_tensors, train_targets, 
              validation_data=(valid_tensors, valid_targets),
              epochs=epochs, batch_size=20, callbacks=[checkpointer], verbose=1)
Train on 6680 samples, validate on 835 samples
Epoch 1/50
6660/6680 [============================>.] - ETA: 0s - loss: 3.5976 - acc: 0.1524Epoch 00000: val_loss improved from inf to 3.81821, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 3.5994 - acc: 0.1522 - val_loss: 3.8182 - val_acc: 0.1281
Epoch 2/50
6660/6680 [============================>.] - ETA: 0s - loss: 3.4692 - acc: 0.1707Epoch 00001: val_loss improved from 3.81821 to 3.57400, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 3.4684 - acc: 0.1707 - val_loss: 3.5740 - val_acc: 0.1581
Epoch 3/50
6660/6680 [============================>.] - ETA: 0s - loss: 3.3335 - acc: 0.1971Epoch 00002: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 3.3346 - acc: 0.1970 - val_loss: 3.5955 - val_acc: 0.1605
Epoch 4/50
6660/6680 [============================>.] - ETA: 0s - loss: 3.1937 - acc: 0.2188Epoch 00003: val_loss improved from 3.57400 to 3.46452, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 3.1944 - acc: 0.2189 - val_loss: 3.4645 - val_acc: 0.1737
Epoch 5/50
6660/6680 [============================>.] - ETA: 0s - loss: 3.0802 - acc: 0.2413Epoch 00004: val_loss improved from 3.46452 to 3.38332, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 3.0804 - acc: 0.2410 - val_loss: 3.3833 - val_acc: 0.2096
Epoch 6/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.9636 - acc: 0.2698Epoch 00005: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 2.9626 - acc: 0.2696 - val_loss: 3.4822 - val_acc: 0.2216
Epoch 7/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.8369 - acc: 0.2946Epoch 00006: val_loss improved from 3.38332 to 3.25498, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 2.8362 - acc: 0.2948 - val_loss: 3.2550 - val_acc: 0.2563
Epoch 8/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.7360 - acc: 0.3059Epoch 00007: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 2.7354 - acc: 0.3060 - val_loss: 3.4736 - val_acc: 0.1928
Epoch 9/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.6232 - acc: 0.3381Epoch 00008: val_loss improved from 3.25498 to 3.12825, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 2.6242 - acc: 0.3380 - val_loss: 3.1282 - val_acc: 0.2563
Epoch 10/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.4983 - acc: 0.3631Epoch 00009: val_loss improved from 3.12825 to 3.08868, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 2.5007 - acc: 0.3626 - val_loss: 3.0887 - val_acc: 0.2683
Epoch 11/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.4181 - acc: 0.3757Epoch 00010: val_loss improved from 3.08868 to 2.98806, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 2.4196 - acc: 0.3753 - val_loss: 2.9881 - val_acc: 0.2970
Epoch 12/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.3165 - acc: 0.3980Epoch 00011: val_loss improved from 2.98806 to 2.94160, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 2.3158 - acc: 0.3981 - val_loss: 2.9416 - val_acc: 0.3234
Epoch 13/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.2173 - acc: 0.4252Epoch 00012: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 2.2180 - acc: 0.4251 - val_loss: 3.1377 - val_acc: 0.2431
Epoch 14/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.1455 - acc: 0.4392Epoch 00013: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 2.1448 - acc: 0.4391 - val_loss: 2.9505 - val_acc: 0.3162
Epoch 15/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.0477 - acc: 0.4620Epoch 00014: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 2.0472 - acc: 0.4620 - val_loss: 3.2609 - val_acc: 0.3090
Epoch 16/50
6660/6680 [============================>.] - ETA: 0s - loss: 2.0117 - acc: 0.4697Epoch 00015: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 2.0106 - acc: 0.4696 - val_loss: 2.9544 - val_acc: 0.3389
Epoch 17/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.9041 - acc: 0.4940Epoch 00016: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.9042 - acc: 0.4942 - val_loss: 3.0061 - val_acc: 0.3102
Epoch 18/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.8207 - acc: 0.5111Epoch 00017: val_loss improved from 2.94160 to 2.88123, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 1.8212 - acc: 0.5115 - val_loss: 2.8812 - val_acc: 0.3353
Epoch 19/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.7546 - acc: 0.5200Epoch 00018: val_loss improved from 2.88123 to 2.79554, saving model to saved_models/weights.best.from_scratch.hdf5
6680/6680 [==============================] - 56s - loss: 1.7553 - acc: 0.5199 - val_loss: 2.7955 - val_acc: 0.3605
Epoch 20/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.6903 - acc: 0.5375Epoch 00019: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.6885 - acc: 0.5380 - val_loss: 3.0758 - val_acc: 0.3629
Epoch 21/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.6495 - acc: 0.5497Epoch 00020: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.6496 - acc: 0.5493 - val_loss: 3.0021 - val_acc: 0.3593
Epoch 22/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.5417 - acc: 0.5733Epoch 00021: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.5415 - acc: 0.5735 - val_loss: 2.9111 - val_acc: 0.3808
Epoch 23/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.5126 - acc: 0.5832Epoch 00022: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.5132 - acc: 0.5828 - val_loss: 3.1137 - val_acc: 0.3150
Epoch 24/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.4483 - acc: 0.5998Epoch 00023: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.4492 - acc: 0.5996 - val_loss: 3.0748 - val_acc: 0.3150
Epoch 25/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.4031 - acc: 0.6021Epoch 00024: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.4058 - acc: 0.6021 - val_loss: 3.5325 - val_acc: 0.3341
Epoch 26/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.3631 - acc: 0.6195Epoch 00025: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.3652 - acc: 0.6190 - val_loss: 3.0017 - val_acc: 0.3449
Epoch 27/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.3210 - acc: 0.6287Epoch 00026: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.3195 - acc: 0.6289 - val_loss: 3.0701 - val_acc: 0.3713
Epoch 28/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.2794 - acc: 0.6444Epoch 00027: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.2829 - acc: 0.6434 - val_loss: 2.9578 - val_acc: 0.3677
Epoch 29/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.2311 - acc: 0.6521Epoch 00028: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.2327 - acc: 0.6518 - val_loss: 3.0577 - val_acc: 0.3353
Epoch 30/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.1949 - acc: 0.6703Epoch 00029: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.1931 - acc: 0.6707 - val_loss: 2.9794 - val_acc: 0.4156
Epoch 31/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.1635 - acc: 0.6716Epoch 00030: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.1658 - acc: 0.6708 - val_loss: 5.8154 - val_acc: 0.2383
Epoch 32/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.1611 - acc: 0.6691Epoch 00031: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.1595 - acc: 0.6696 - val_loss: 2.9901 - val_acc: 0.3844
Epoch 33/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.1150 - acc: 0.6902Epoch 00032: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.1161 - acc: 0.6898 - val_loss: 3.2943 - val_acc: 0.3377
Epoch 34/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.0999 - acc: 0.6898Epoch 00033: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.1003 - acc: 0.6895 - val_loss: 3.5148 - val_acc: 0.3701
Epoch 35/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.0404 - acc: 0.6979Epoch 00034: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.0395 - acc: 0.6981 - val_loss: 3.2510 - val_acc: 0.3725
Epoch 36/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.0579 - acc: 0.7056Epoch 00035: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.0589 - acc: 0.7054 - val_loss: 3.0158 - val_acc: 0.4060
Epoch 37/50
6660/6680 [============================>.] - ETA: 0s - loss: 1.0151 - acc: 0.7066Epoch 00036: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 1.0127 - acc: 0.7073 - val_loss: 3.1509 - val_acc: 0.4180
Epoch 38/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9762 - acc: 0.7246Epoch 00037: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9763 - acc: 0.7247 - val_loss: 3.3797 - val_acc: 0.3856
Epoch 39/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9567 - acc: 0.7302Epoch 00038: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9560 - acc: 0.7304 - val_loss: 3.5132 - val_acc: 0.3665
Epoch 40/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9669 - acc: 0.7288Epoch 00039: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9684 - acc: 0.7284 - val_loss: 5.0049 - val_acc: 0.2659
Epoch 41/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9331 - acc: 0.7365Epoch 00040: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9326 - acc: 0.7365 - val_loss: 3.6146 - val_acc: 0.3976
Epoch 42/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9344 - acc: 0.7383Epoch 00041: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9352 - acc: 0.7382 - val_loss: 3.0375 - val_acc: 0.4000
Epoch 43/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9522 - acc: 0.7387Epoch 00042: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9530 - acc: 0.7385 - val_loss: 3.3338 - val_acc: 0.3988
Epoch 44/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9145 - acc: 0.7545Epoch 00043: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9171 - acc: 0.7543 - val_loss: 3.9268 - val_acc: 0.2838
Epoch 45/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9148 - acc: 0.7500Epoch 00044: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9144 - acc: 0.7499 - val_loss: 3.4665 - val_acc: 0.3820
Epoch 46/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.8877 - acc: 0.7465Epoch 00045: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.8903 - acc: 0.7458 - val_loss: 3.4613 - val_acc: 0.3509
Epoch 47/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9020 - acc: 0.7538Epoch 00046: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9030 - acc: 0.7531 - val_loss: 4.1017 - val_acc: 0.3030
Epoch 48/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.8979 - acc: 0.7568Epoch 00047: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.8978 - acc: 0.7567 - val_loss: 3.2086 - val_acc: 0.4012
Epoch 49/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.8967 - acc: 0.7556Epoch 00048: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.8944 - acc: 0.7561 - val_loss: 3.0897 - val_acc: 0.4228
Epoch 50/50
6660/6680 [============================>.] - ETA: 0s - loss: 0.9216 - acc: 0.7556Epoch 00049: val_loss did not improve
6680/6680 [==============================] - 56s - loss: 0.9271 - acc: 0.7549 - val_loss: 3.2824 - val_acc: 0.3904
In [20]:
import matplotlib.pyplot as plt
%matplotlib inline

# summarize history for accuracy
plt.plot(history.history['acc'])
plt.plot(history.history['val_acc'])
plt.title('model accuracy')
plt.ylabel('accuracy')
plt.xlabel('epoch')
plt.legend(['train', 'valid'], loc='upper left')
plt.show()

# summarize history for loss
plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'valid'], loc='upper left')
plt.show()
In [21]:
from keras import backend as K

# util function to convert a tensor into a valid image
def deprocess_image(x):
    # normalize tensor: center on 0., ensure std is 0.1
    x -= x.mean()
    x /= (x.std() + 1e-5)
    x *= 0.1

    # clip to [0, 1]
    x += 0.5
    x = np.clip(x, 0, 1)

    # convert to RGB array
    x *= 255
    
    if K.image_dim_ordering() == 'th':
        x = x.transpose((1, 2, 0))

    x = np.clip(x, 0, 255).astype('uint8')
    return x

def normalize(x):
    # utility function to normalize a tensor by its L2 norm
    return x / (K.sqrt(K.mean(K.square(x))) + K.epsilon())

def get_filters_img(filters, img_width, img_height):
    margin = 5
    n = int(len(filters)**0.5)
    width = n * img_width + (n - 1) * margin
    height = n * img_height + (n - 1) * margin
    stitched_filters = np.zeros((width, height, 3))

    # fill the picture with our saved filters
    for i in range(n):
        for j in range(n):
            index = i * n + j
            if index < len(filters):
                img = filters[i * n + j]
                stitched_filters[(img_width + margin) * i: (img_width + margin) * i + img_width,
                                 (img_height + margin) * j: (img_height + margin) * j + img_height, :] = img

    return stitched_filters
In [22]:
from matplotlib.pyplot import imshow
%matplotlib inline

img_width = 224
img_height = 224

input_img = image.load_img(train_files[0], target_size=(img_width, img_height))
input_img = np.asarray(input_img)
imshow(input_img)
Out[22]:
<matplotlib.image.AxesImage at 0x7f625a6a7470>
In [23]:
import math
import time

def show_layer_output(layer_name, filters_num, img_width=224, img_height=224):
    input_placeholder = model.input
    kept_filters = []

    for filter_index in range(filters_num):
        # we build a loss function that maximizes the activation
        # of the nth filter of the layer considered
        layer_output = layer_dict[layer_name].output
        if K.image_data_format() == 'channels_first':
            loss = K.mean(layer_output[:, filter_index, :, :])
        else:
            loss = K.mean(layer_output[:, :, :, filter_index])

        # we compute the gradient of the input picture with this loss
        grads = K.gradients(loss, input_placeholder)[0]

        # normalization trick: we normalize the gradient
        grads = normalize(grads)

        # this function returns the loss and grads given the input picture
        iterate = K.function([input_placeholder], [loss, grads])

        # step size for gradient ascent
        step = 1.

        '''
        # we start from a sample image from the training set
        if K.image_data_format() == 'channels_first':
            input_img_data = np.asarray([np.transpose(input_img, (2, 0, 1))], dtype='float32')
        else:
            input_img_data = np.asarray([input_img], dtype='float32')
        '''
        # we start from a gray image with some random noise
        if K.image_data_format() == 'channels_first':
            input_img_data = np.random.random((1, 3, img_width, img_height))
        else:
            input_img_data = np.random.random((1, img_width, img_height, 3))
        input_img_data = (input_img_data - 0.5) * 20 + 128


        # we run gradient ascent for 20 steps
        for i in range(20):
            loss_value, grads_value = iterate([input_img_data])
            input_img_data += grads_value * step

            if loss_value <= 0.:
                # some filters get stuck to 0, we can skip them
                break

        # decode the resulting input image
        if loss_value > 0:
            img = deprocess_image(input_img_data[0])
            kept_filters.append((img, loss_value))
        end_time = time.time()

    # we will stich the best 64 filters on a 8 x 8 grid.
    n = 8
    kept_size = kept_max = n * n

    # the filters that have the highest loss are assumed to be better-looking.
    kept_filters.sort(key=lambda x: x[1], reverse=True)
    if len(kept_filters) > kept_max:
        kept_filters = kept_filters[:kept_max]
    else:
        kept_size = len(kept_filters)

    # build a black picture with enough space for
    # our 8 x 8 filters of size 224 x 224, with a 5px margin in between
    margin = 5
    width = n * img_width + (n - 1) * margin
    row_cnt = math.ceil(kept_size / n)
    height = row_cnt * img_height + (row_cnt - 1) * margin
    stitched_filters = np.zeros((height, width, 3))

    # fill the picture with our saved filters
    kept_idx = 0
    for i in range(row_cnt):
        for j in range(n):
            if kept_idx >= kept_size:
                break

            img, loss = kept_filters[kept_idx]
            kept_idx += 1

            stitched_filters[(img_height + margin) * i: (img_height + margin) * i + img_height,
                             (img_width + margin) * j: (img_width + margin) * j + img_width, :] = img
    return stitched_filters
In [24]:
# filters_num = iter([16, 32, 64, 128, 256])
filters_num = iter([32, 64, 128, 256, 512])
layer_dict = dict([(layer.name, layer) for layer in model.layers])
for key in layer_dict:
    if key.startswith('conv2d_'):
        fig = plt.figure(figsize=(32, 32))
        plt.title(key)
        imshow(show_layer_output(key, next(filters_num)), interpolation='none')

Load the Model with the Best Validation Loss

In [25]:
model.load_weights('saved_models/weights.best.from_scratch.hdf5')

Test the Model

Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 1%.

In [26]:
# get index of predicted dog breed for each image in test set
dog_breed_predictions = [np.argmax(model.predict(np.expand_dims(tensor, axis=0))) for tensor in test_tensors]

# report test accuracy
test_accuracy = 100*np.sum(np.array(dog_breed_predictions)==np.argmax(test_targets, axis=1))/len(dog_breed_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 36.0048%

Step 4: Use a CNN to Classify Dog Breeds

To reduce training time without sacrificing accuracy, we show you how to train a CNN using transfer learning. In the following step, you will get a chance to use transfer learning to train your own CNN.

Obtain Bottleneck Features

In [27]:
bottleneck_features = np.load('bottleneck_features/DogVGG16Data.npz')
train_VGG16 = bottleneck_features['train']
valid_VGG16 = bottleneck_features['valid']
test_VGG16 = bottleneck_features['test']

Model Architecture

The model uses the the pre-trained VGG-16 model as a fixed feature extractor, where the last convolutional output of VGG-16 is fed as input to our model. We only add a global average pooling layer and a fully connected layer, where the latter contains one node for each dog category and is equipped with a softmax.

In [28]:
VGG16_model = Sequential()
VGG16_model.add(GlobalAveragePooling2D(input_shape=train_VGG16.shape[1:]))
VGG16_model.add(Dense(133, activation='softmax'))

VGG16_model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
global_average_pooling2d_2 ( (None, 512)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 133)               68229     
=================================================================
Total params: 68,229
Trainable params: 68,229
Non-trainable params: 0
_________________________________________________________________

Compile the Model

In [29]:
VGG16_model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

Train the Model

In [30]:
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.VGG16.hdf5', 
                               verbose=1, save_best_only=True)

VGG16_model.fit(train_VGG16, train_targets, 
          validation_data=(valid_VGG16, valid_targets),
          epochs=20, batch_size=20, callbacks=[checkpointer], verbose=1)
Train on 6680 samples, validate on 835 samples
Epoch 1/20
6380/6680 [===========================>..] - ETA: 0s - loss: 12.7911 - acc: 0.1041Epoch 00000: val_loss improved from inf to 11.53865, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 2s - loss: 12.7098 - acc: 0.1088 - val_loss: 11.5386 - val_acc: 0.1725
Epoch 2/20
6480/6680 [============================>.] - ETA: 0s - loss: 10.6313 - acc: 0.2574Epoch 00001: val_loss improved from 11.53865 to 10.68676, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 10.6472 - acc: 0.2569 - val_loss: 10.6868 - val_acc: 0.2455
Epoch 3/20
6520/6680 [============================>.] - ETA: 0s - loss: 10.1864 - acc: 0.3133Epoch 00002: val_loss improved from 10.68676 to 10.46578, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 10.1922 - acc: 0.3136 - val_loss: 10.4658 - val_acc: 0.2743
Epoch 4/20
6500/6680 [============================>.] - ETA: 0s - loss: 9.9944 - acc: 0.3388 Epoch 00003: val_loss improved from 10.46578 to 10.30551, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.9661 - acc: 0.3409 - val_loss: 10.3055 - val_acc: 0.2934
Epoch 5/20
6480/6680 [============================>.] - ETA: 0s - loss: 9.7052 - acc: 0.3596Epoch 00004: val_loss improved from 10.30551 to 10.05498, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.6768 - acc: 0.3611 - val_loss: 10.0550 - val_acc: 0.3054
Epoch 6/20
6500/6680 [============================>.] - ETA: 0s - loss: 9.5063 - acc: 0.3823Epoch 00005: val_loss improved from 10.05498 to 9.99985, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.5368 - acc: 0.3805 - val_loss: 9.9999 - val_acc: 0.3126
Epoch 7/20
6480/6680 [============================>.] - ETA: 0s - loss: 9.3689 - acc: 0.3934Epoch 00006: val_loss improved from 9.99985 to 9.84984, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.3537 - acc: 0.3939 - val_loss: 9.8498 - val_acc: 0.3246
Epoch 8/20
6520/6680 [============================>.] - ETA: 0s - loss: 9.1399 - acc: 0.4109Epoch 00007: val_loss improved from 9.84984 to 9.59475, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 9.1232 - acc: 0.4117 - val_loss: 9.5948 - val_acc: 0.3449
Epoch 9/20
6540/6680 [============================>.] - ETA: 0s - loss: 8.9796 - acc: 0.4220Epoch 00008: val_loss improved from 9.59475 to 9.52706, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.9788 - acc: 0.4223 - val_loss: 9.5271 - val_acc: 0.3425
Epoch 10/20
6580/6680 [============================>.] - ETA: 0s - loss: 8.9156 - acc: 0.4283Epoch 00009: val_loss improved from 9.52706 to 9.40621, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.8981 - acc: 0.4296 - val_loss: 9.4062 - val_acc: 0.3425
Epoch 11/20
6520/6680 [============================>.] - ETA: 0s - loss: 8.6402 - acc: 0.4399Epoch 00010: val_loss improved from 9.40621 to 9.29619, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.6261 - acc: 0.4401 - val_loss: 9.2962 - val_acc: 0.3425
Epoch 12/20
6580/6680 [============================>.] - ETA: 0s - loss: 8.4197 - acc: 0.4591Epoch 00011: val_loss improved from 9.29619 to 9.09836, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.4322 - acc: 0.4582 - val_loss: 9.0984 - val_acc: 0.3557
Epoch 13/20
6540/6680 [============================>.] - ETA: 0s - loss: 8.2911 - acc: 0.4685Epoch 00012: val_loss improved from 9.09836 to 8.97059, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.2949 - acc: 0.4684 - val_loss: 8.9706 - val_acc: 0.3689
Epoch 14/20
6560/6680 [============================>.] - ETA: 0s - loss: 8.2287 - acc: 0.4782Epoch 00013: val_loss improved from 8.97059 to 8.85165, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.2337 - acc: 0.4778 - val_loss: 8.8516 - val_acc: 0.3856
Epoch 15/20
6500/6680 [============================>.] - ETA: 0s - loss: 8.1600 - acc: 0.4849Epoch 00014: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 8.1767 - acc: 0.4840 - val_loss: 8.8772 - val_acc: 0.3976
Epoch 16/20
6400/6680 [===========================>..] - ETA: 0s - loss: 8.0738 - acc: 0.4867Epoch 00015: val_loss improved from 8.85165 to 8.63911, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 8.0428 - acc: 0.4889 - val_loss: 8.6391 - val_acc: 0.3988
Epoch 17/20
6360/6680 [===========================>..] - ETA: 0s - loss: 7.8261 - acc: 0.4970Epoch 00016: val_loss improved from 8.63911 to 8.40572, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.8208 - acc: 0.4973 - val_loss: 8.4057 - val_acc: 0.3988
Epoch 18/20
6400/6680 [===========================>..] - ETA: 0s - loss: 7.5096 - acc: 0.5167Epoch 00017: val_loss improved from 8.40572 to 8.31122, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.5129 - acc: 0.5171 - val_loss: 8.3112 - val_acc: 0.4180
Epoch 19/20
6320/6680 [===========================>..] - ETA: 0s - loss: 7.4654 - acc: 0.5280Epoch 00018: val_loss improved from 8.31122 to 8.26795, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.4566 - acc: 0.5290 - val_loss: 8.2680 - val_acc: 0.4180
Epoch 20/20
6380/6680 [===========================>..] - ETA: 0s - loss: 7.4272 - acc: 0.5332Epoch 00019: val_loss improved from 8.26795 to 8.19371, saving model to saved_models/weights.best.VGG16.hdf5
6680/6680 [==============================] - 0s - loss: 7.4402 - acc: 0.5319 - val_loss: 8.1937 - val_acc: 0.4216
Out[30]:
<keras.callbacks.History at 0x7f60ca848f60>

Load the Model with the Best Validation Loss

In [31]:
VGG16_model.load_weights('saved_models/weights.best.VGG16.hdf5')

Test the Model

Now, we can use the CNN to test how well it identifies breed within our test dataset of dog images. We print the test accuracy below.

In [32]:
# get index of predicted dog breed for each image in test set
VGG16_predictions = [np.argmax(VGG16_model.predict(np.expand_dims(feature, axis=0))) for feature in test_VGG16]

# report test accuracy
test_accuracy = 100*np.sum(np.array(VGG16_predictions)==np.argmax(test_targets, axis=1))/len(VGG16_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 42.9426%

Predict Dog Breed with the Model

In [33]:
from extract_bottleneck_features import *

def VGG16_predict_breed(img_path):
    # extract bottleneck features
    bottleneck_feature = extract_VGG16(path_to_tensor(img_path))
    # obtain predicted vector
    predicted_vector = VGG16_model.predict(bottleneck_feature)
    # return dog breed that is predicted by the model
    return dog_names[np.argmax(predicted_vector)]

Step 5: Create a CNN to Classify Dog Breeds (using Transfer Learning)

You will now use transfer learning to create a CNN that can identify dog breed from images. Your CNN must attain at least 60% accuracy on the test set.

In Step 4, we used transfer learning to create a CNN using VGG-16 bottleneck features. In this section, you must use the bottleneck features from a different pre-trained model. To make things easier for you, we have pre-computed the features for all of the networks that are currently available in Keras:

The files are encoded as such:

Dog{network}Data.npz

where {network}, in the above filename, can be one of VGG19, Resnet50, InceptionV3, or Xception. Pick one of the above architectures, download the corresponding bottleneck features, and store the downloaded file in the bottleneck_features/ folder in the repository.

(IMPLEMENTATION) Obtain Bottleneck Features

In the code block below, extract the bottleneck features corresponding to the train, test, and validation sets by running the following:

bottleneck_features = np.load('bottleneck_features/Dog{network}Data.npz')
train_{network} = bottleneck_features['train']
valid_{network} = bottleneck_features['valid']
test_{network} = bottleneck_features['test']
In [34]:
### TODO: Obtain bottleneck features from another pre-trained CNN.
bottleneck_features = np.load('bottleneck_features/DogResnet50Data.npz')
train_Resnet50 = bottleneck_features['train']
valid_Resnet50 = bottleneck_features['valid']
test_Resnet50 = bottleneck_features['test']

(IMPLEMENTATION) Model Architecture

Create a CNN to classify dog breed. At the end of your code cell block, summarize the layers of your model by executing the line:

    <your model's name>.summary()

Question 5: Outline the steps you took to get to your final CNN architecture and your reasoning at each step. Describe why you think the architecture is suitable for the current problem.

Answer: I reused all 'DogResnet50Data' features (weights and biases) for convolutional layers. I only gathered them using 'GlobalAveragePooling2D' and made a fully connected layer to classify 133 dog breeds.

In [35]:
### TODO: Define your architecture.
Resnet50_model = Sequential()
Resnet50_model.add(GlobalAveragePooling2D(input_shape=train_Resnet50.shape[1:]))
Resnet50_model.add(Dense(133, activation='softmax'))

Resnet50_model.summary()
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
global_average_pooling2d_3 ( (None, 2048)              0         
_________________________________________________________________
dense_3 (Dense)              (None, 133)               272517    
=================================================================
Total params: 272,517
Trainable params: 272,517
Non-trainable params: 0
_________________________________________________________________

(IMPLEMENTATION) Compile the Model

In [36]:
### TODO: Compile the model.
Resnet50_model.compile(loss='categorical_crossentropy', optimizer='rmsprop', metrics=['accuracy'])

(IMPLEMENTATION) Train the Model

Train your model in the code cell below. Use model checkpointing to save the model that attains the best validation loss.

You are welcome to augment the training data, but this is not a requirement.

In [37]:
### TODO: Train the model.
checkpointer = ModelCheckpoint(filepath='saved_models/weights.best.Resnet50.hdf5', 
                               verbose=1, save_best_only=True)

Resnet50_model.fit(train_Resnet50, train_targets, 
          validation_data=(valid_Resnet50, valid_targets),
          epochs=20, batch_size=20, callbacks=[checkpointer], verbose=1)
Train on 6680 samples, validate on 835 samples
Epoch 1/20
6520/6680 [============================>.] - ETA: 0s - loss: 1.6534 - acc: 0.5933Epoch 00000: val_loss improved from inf to 0.76298, saving model to saved_models/weights.best.Resnet50.hdf5
6680/6680 [==============================] - 2s - loss: 1.6321 - acc: 0.5976 - val_loss: 0.7630 - val_acc: 0.7521
Epoch 2/20
6600/6680 [============================>.] - ETA: 0s - loss: 0.4399 - acc: 0.8615Epoch 00001: val_loss improved from 0.76298 to 0.67984, saving model to saved_models/weights.best.Resnet50.hdf5
6680/6680 [==============================] - 0s - loss: 0.4389 - acc: 0.8617 - val_loss: 0.6798 - val_acc: 0.7892
Epoch 3/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.2610 - acc: 0.9215Epoch 00002: val_loss improved from 0.67984 to 0.67196, saving model to saved_models/weights.best.Resnet50.hdf5
6680/6680 [==============================] - 0s - loss: 0.2601 - acc: 0.9214 - val_loss: 0.6720 - val_acc: 0.7928
Epoch 4/20
6600/6680 [============================>.] - ETA: 0s - loss: 0.1778 - acc: 0.9432Epoch 00003: val_loss improved from 0.67196 to 0.64025, saving model to saved_models/weights.best.Resnet50.hdf5
6680/6680 [==============================] - 0s - loss: 0.1775 - acc: 0.9431 - val_loss: 0.6403 - val_acc: 0.8084
Epoch 5/20
6640/6680 [============================>.] - ETA: 0s - loss: 0.1250 - acc: 0.9592Epoch 00004: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.1256 - acc: 0.9590 - val_loss: 0.6485 - val_acc: 0.8108
Epoch 6/20
6600/6680 [============================>.] - ETA: 0s - loss: 0.0831 - acc: 0.9762Epoch 00005: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0830 - acc: 0.9765 - val_loss: 0.7045 - val_acc: 0.8120
Epoch 7/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0675 - acc: 0.9792Epoch 00006: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0678 - acc: 0.9789 - val_loss: 0.7180 - val_acc: 0.8096
Epoch 8/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0485 - acc: 0.9864Epoch 00007: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0482 - acc: 0.9865 - val_loss: 0.6929 - val_acc: 0.8012
Epoch 9/20
6640/6680 [============================>.] - ETA: 0s - loss: 0.0360 - acc: 0.9902Epoch 00008: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0359 - acc: 0.9903 - val_loss: 0.6823 - val_acc: 0.8228
Epoch 10/20
6660/6680 [============================>.] - ETA: 0s - loss: 0.0283 - acc: 0.9931Epoch 00009: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0283 - acc: 0.9931 - val_loss: 0.7833 - val_acc: 0.8144
Epoch 11/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0217 - acc: 0.9952Epoch 00010: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0220 - acc: 0.9949 - val_loss: 0.7774 - val_acc: 0.8180
Epoch 12/20
6600/6680 [============================>.] - ETA: 0s - loss: 0.0181 - acc: 0.9955Epoch 00011: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0181 - acc: 0.9955 - val_loss: 0.7986 - val_acc: 0.8144
Epoch 13/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0150 - acc: 0.9965Epoch 00012: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0149 - acc: 0.9966 - val_loss: 0.7877 - val_acc: 0.8120
Epoch 14/20
6640/6680 [============================>.] - ETA: 0s - loss: 0.0126 - acc: 0.9976Epoch 00013: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0126 - acc: 0.9976 - val_loss: 0.7947 - val_acc: 0.8371
Epoch 15/20
6640/6680 [============================>.] - ETA: 0s - loss: 0.0108 - acc: 0.9970Epoch 00014: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0108 - acc: 0.9970 - val_loss: 0.8120 - val_acc: 0.8263
Epoch 16/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0109 - acc: 0.9977Epoch 00015: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0108 - acc: 0.9978 - val_loss: 0.8916 - val_acc: 0.8084
Epoch 17/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0084 - acc: 0.9983Epoch 00016: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0083 - acc: 0.9984 - val_loss: 0.8682 - val_acc: 0.8204
Epoch 18/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0077 - acc: 0.9979Epoch 00017: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0076 - acc: 0.9979 - val_loss: 0.8862 - val_acc: 0.8263
Epoch 19/20
6620/6680 [============================>.] - ETA: 0s - loss: 0.0069 - acc: 0.9982Epoch 00018: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0068 - acc: 0.9982 - val_loss: 0.8843 - val_acc: 0.8240
Epoch 20/20
6640/6680 [============================>.] - ETA: 0s - loss: 0.0055 - acc: 0.9983Epoch 00019: val_loss did not improve
6680/6680 [==============================] - 0s - loss: 0.0055 - acc: 0.9984 - val_loss: 0.9008 - val_acc: 0.8216
Out[37]:
<keras.callbacks.History at 0x7f60ca5eb5c0>

(IMPLEMENTATION) Load the Model with the Best Validation Loss

In [45]:
### TODO: Load the model weights with the best validation loss.
Resnet50_model.load_weights('saved_models/weights.best.Resnet50.hdf5')

(IMPLEMENTATION) Test the Model

Try out your model on the test dataset of dog images. Ensure that your test accuracy is greater than 60%.

In [46]:
### TODO: Calculate classification accuracy on the test dataset.
# get index of predicted dog breed for each image in test set
Resnet50_predictions = [np.argmax(Resnet50_model.predict(np.expand_dims(feature, axis=0))) for feature in test_Resnet50]

# report test accuracy
test_accuracy = 100*np.sum(np.array(Resnet50_predictions)==np.argmax(test_targets, axis=1))/len(Resnet50_predictions)
print('Test accuracy: %.4f%%' % test_accuracy)
Test accuracy: 81.5789%

(IMPLEMENTATION) Predict Dog Breed with the Model

Write a function that takes an image path as input and returns the dog breed (Affenpinscher, Afghan_hound, etc) that is predicted by your model.

Similar to the analogous function in Step 5, your function should have three steps:

  1. Extract the bottleneck features corresponding to the chosen CNN model.
  2. Supply the bottleneck features as input to the model to return the predicted vector. Note that the argmax of this prediction vector gives the index of the predicted dog breed.
  3. Use the dog_names array defined in Step 0 of this notebook to return the corresponding breed.

The functions to extract the bottleneck features can be found in extract_bottleneck_features.py, and they have been imported in an earlier code cell. To obtain the bottleneck features corresponding to your chosen CNN architecture, you need to use the function

extract_{network}

where {network}, in the above filename, should be one of VGG19, Resnet50, InceptionV3, or Xception.

In [47]:
### TODO: Write a function that takes a path to an image as input
### and returns the dog breed that is predicted by the model.
from extract_bottleneck_features import *

def Resnet50_predict_breed(img_path):
    # extract bottleneck features
    bottleneck_feature = extract_Resnet50(path_to_tensor(img_path))
    # obtain predicted vector
    predicted_vector = Resnet50_model.predict(bottleneck_feature)
    # return dog breed that is predicted by the model
    return dog_names[np.argmax(predicted_vector)]

Step 6: Write your Algorithm

Write an algorithm that accepts a file path to an image and first determines whether the image contains a human, dog, or neither. Then,

  • if a dog is detected in the image, return the predicted breed.
  • if a human is detected in the image, return the resembling dog breed.
  • if neither is detected in the image, provide output that indicates an error.

You are welcome to write your own functions for detecting humans and dogs in images, but feel free to use the face_detector and dog_detector functions developed above. You are required to use your CNN from Step 5 to predict dog breed.

Some sample output for our algorithm is provided below, but feel free to design your own user experience!

Sample Human Output

(IMPLEMENTATION) Write your Algorithm

In [48]:
### TODO: Write your algorithm.
### Feel free to use as many code cells as needed.
def find_dog_breed(img_path):
    pos, prediction = dog_detector(img_path)
    if pos == True:
        return True, Resnet50_predict_breed(img_path)
    
    pos, _ = face_detector(img_path)
    if pos == True:
        return True, 'You look like a ...\n' + Resnet50_predict_breed(img_path)
    
    return False, ''

Step 7: Test Your Algorithm

In this section, you will take your new algorithm for a spin! What kind of dog does the algorithm think that you look like? If you have a dog, does it predict your dog's breed accurately? If you have a cat, does it mistakenly think that your cat is a dog?

(IMPLEMENTATION) Test Your Algorithm on Sample Images!

Test your algorithm at least six images on your computer. Feel free to use any images you like. Use at least two human and two dog images.

Question 6: Is the output better than you expected :) ? Or worse :( ? Provide at least three possible points of improvement for your algorithm.

Answer:

  • 'dog_detector' is not perfect. Showing the location of the dog in the image will be helpful to find out the problem.
  • Instead of using 'face_detector' and 'dog_detector', use a new CNN to classify dog or human.
  • If an image has two or more dogs, humans, or mix, run multiple independent CNN to detect dogs and humans simultaneously. Or, find locations of dogs and humans at first.
In [60]:
import matplotlib.image as mpimg

def show_img(img_path):
    img = mpimg.imread(img_path)
    pos, breed = find_dog_breed(img_path)
    if pos == True:
        plt.title(breed)
    else:
        plt.title('Unknown Breed')
    plt.imshow(img)
    plt.show()

## TODO: Execute your algorithm from Step 6 on
## at least 6 images on your computer.
## Feel free to use as many code cells as needed.
show_img('images/i.jpeg')
show_img('images/images.jpeg')
show_img('images/o-PUPPIES-570.jpg')
show_img('images/IMG_0743.JPG')
show_img('images/IMG_0766.JPG')
show_img('images/IMG_0874.JPG')
show_img('images/IMG_0707.JPG')
show_img('images/IMG_0753.JPG')
show_img('images/IMG_0780.JPG')
show_img('images/Old-Man-With-Old-Dog-Funny-Picture.jpg')